> ... because I set mine up with cpanm

1) I for sure will give this way a try. I had not the best experience with this tool but if it offers a cleaner, inteleggible, customizable and easy way to set up a cpan tester environment I will adopt it without esitations.

My intention is to spot all dark corners of the cpan tester setup in strawberry and, possibly, end with a little walktrough: infact i see few win32 testers and many times using punch card era machines (my respect for them anyway but the spectrum is by far too little).

> cpan-specific questions -- like the uninitialized errors

2) This is one of my major dissapoint at the moment. I'm used to see perl running smoothly and shining for its intrinsic quality. I expect (with all the patience I'm able of) the toolchain to be super portable, super tested, super simple, super reliable. Super as perl. Which modules of the dozens installed by the Task::CPAN::Reporter bundle causes the error? Not even using Carp?

Little excursus: why this huge amount of non core dependencies? I see this signaled as issue and srezic propose a super simple approach I very like this direction: cut, cut, cut.. I'll try his way too.

> I'm assuming these are the tests you were mentioning: I now see three, two with no name, and one with the name present. If so, your most-recent submission used your name.

3) Yes they are, but I'm conducing various tests from two machines and always with brand new strawberry installations with a sanitized PATH and the third report comes from my home machine (and see also my note at step 4 above). I'm collecting all quirks and unwanted behaviours and probably results from this other test will go in a sperate post in this thread.

> However, I do know that I've successfully copied my .cpanreporter folder to multiple versions of portable strawberry

4) Theese are good news! Which steps you'have done before copying it? Also this must be documented in a better way, imho.

> your PAUSE ID here, and I remembered your sadness a few days back of so few passes

5) ;=) thanks! you are very kind even if this is the very last of my concerns at the moment. The whole point of my experimental module is the presence of an executable that I supposed to be present almost since ever, but were not present in these punch card machine of the respected testers. No worry ;) I have added a check in the Makefile.PL and, even if not optimal I know the problem. I also have few things to change. i call it an experiment because was my own gymnasium confronting my knoledges in the wild. But thanks!

L*

>>... because I set mine up with cpanm

> 1) I for sure will give this way a try. I had not the best experience with this tool but if it offers a cleaner, inteleggible, customizable and easy way to set up a cpan tester environment I will adopt it without esitations.

I don't know that I'd go so far as "cleaner". I just did the cpanm version, because I always use cpanm to install modules already. One of the downfalls is that cpanm-reporter does not auto-report after installing a module, whereas the docs imply that the cpan and cpanplus solutions do. Further, if you're going to use old versions (especially strawberry's 5.10), cpanm is hard to get working right (at least for me), and cpanm-reporter is even harder.

> My intention is to spot all dark corners of the cpan tester setup in strawberry and, possibly, end with a little walktrough: infact i see few win32 testers and many times using punch card era machines (my respect for them anyway but the spectrum is by far too little).

Yeah, the lack of win32 testers is one of the reasons I setup cpanm-reporter for myself, as well. For the "punch card era machines", I think, the issue is from the "smoke testers" -- ie, the ones that set it up to automatically test new modules as they are submitted to CPAN, rather than just on-install testers like myself -- unlike linux, Windows OS aren't free, and a smoke-testing machine is often a dedicated machine, so it's easier to use an older PC, than to buy a machine just to smoke-test, or to try to multitask a machine that's doing smoke-testing in the background, but is your daily-use PC in the foreground. (I also wonder whether many Windows smoke-testers are on virtual machines, and whether those virtual machines might have a limited subset of what's normally installed on modern Windows.)

I really do wish, however, that the Quick Start and related docs would give a better explanation for how to set up an automated smoke-test machine in the Windows/Strawberry environment. I might have or find an older Win-PC that I could dedicate to smoke-testing, if there were a ready-made guide. (And I'd probably be willing to try a non-cpanm solution, if cpanm doesn't lend itself to the automated environment of smoke-testing)

>> However, I do know that I've successfully copied my .cpanreporter folder to multiple versions of portable strawberry

>4) Theese are good news! Which steps you'have done before copying it? Also this must be documented in a better way, imho.

I did the full setup once, similar to what you did in steps1-3, but putting it in the right location for the portable strawberry. I then copied that to a private subversion repo I have. Then, when I do new perl versions that I want to be able to cpanm-reporter, I install App::cpanminus::reporter, then svn export the .cpanreporter directory to the right location, and it's ready to go.

> 5) ;=) thanks! you are very kind even if this is the very last of my concerns at the moment.

That was more of a side-benefit of your post, that I thought I'd let you know about. :-)

I dug up my readme file that I keep with my .cpanreporter repo; it's highly cpanm-specific, but it may come in handy for you (or others), especially if they have a strawberry-v5.10

http://wiki.cpantesters.org/wiki/QuickStart => explains how to get started...
    -- this svn directory includes my metabase_id.json and config.ini...
        QuickStart is a little ambiguous where it goes

I had set this up Fall 2017 (based on dates), but apparently here, that perl install had been lost...

Since I use cpanm, SETUP:
0) per https://github.com/rjbs/Data-UUID/issues/24, bug in Data::UUID install process, so `mkdir c:\tmp`, then `cpanm Data::UUID` first
1) install `cpanm App::cpanminus::reporter`
2) run `cpanm-reporter --setup` (or copy this folder to either berrybrew\perls\#\data\.cpanreporter\ and/or %home%\data\.cpanreporter )

* compatibility notes:
    - 5.10: does not have its own cpanm, so it finds my default-PATH strawberry 5.26 cpanm and tries to use that, which won't work right;
      Workaround = while strawberry 5.10 is selected, 
        * download tarball from "Download" link on https://metacpan.org/pod/cpanm
        * tar xvzf App-cpanminus*.tar.gz
        * cd  App-cpanminusTAB
        * perl Makefile.PL
        * edit Makefile to ensure it doesn't call out c:\strawberry, and instead always uses the specific 5.10 directory
        * dmake
        * #dmake test       # unneeded, because they don't have a real test suite
        * dmake install
        * where cpanm                       # make sure 5.10's cpanm occurs first
        * cpanm --force App::cpanminus      # re-install with cpanm
    - 5.10: at home
        * had to use 5.10.1.2 portable, not berrybrew's 5.10.1.5 zip :-(
        * approximately the following sequence worked for getting reporter installed (there were more atttemps at the main ::reporter package in between):
            cpanm Test::More
            cpanm IO::Prompt::Tiny
            cpanm http://cpan.metacpan.org/authors/id/G/GA/GARU/App-cpanminus-reporter-0.10.tar.gz
            cpanm --notest CPAN::Testers::Common::Client
            cpanm http://cpan.metacpan.org/authors/id/G/GA/GARU/App-cpanminus-reporter-0.10.tar.gz
        * Also having problems with SSL
            try .cpanmreporter/config.ini with http:// instead of https:// => yep, I can send reports now
        * might need to check portable.perl for duplicate 'installbin' or other tags...

Every time I install a package, need to manually run cpanm-reporter afterward...
1) `cpanm Package(s)`
2) `cpanm-reporter`
    -- with my cpanm setup in berrybrew, might have to go into individual %HOME%\.cpanm\work and do `cpanm-reporter --build_dir=.`
    -- oh, maybe it always copies the most recent build.log into %home%\.cpanm\build.log... run some experiments...
        -- 5.12 and 5.16 seemed to update it live
        -- 5.20 seems to copy after the build is completes
        -- 5.26.1_32 -- wouldn't build Data::UUID, even with c:\tmp existing (different error message: 193)
    -- thus, I can use `cpanm-reporter --build_dir=%HOME%\.cpanm` ir
    -- At $work, I have the %HOME% environment variable set (not a default windows variable, but some $work-apps use it);
        * if that's set, then cpanm uses %HOME%\.cpanm...
        * if it's not set (like on my personal home PC), then each portable strawberry hierarchy uses its own .cpanm folder for build.log,
          similar to the own .cpanreporter directory I mentioned above.  using `cpanm --look Some::Distro` will tell you where this .cpanm
          directory is when it opens up a cmd.exe shell in the temporary work directory, which is in the correct .cpanm hierarchy.  
      - I have been known to make a cpanm-with-reporter.bat file, placed next to the cpanm.bat and cpanm-reporter.bat, which contains 
        something like
                @call path\to\cpanm.bat %*
                @call path\to\cpanm-reporter
        
3) check http://metabase.cpantesters.org/tail/log.txt -- should update every 5-10 minutes